arXiv research

PyFi uses adversarial agents to train VLMs on financial image understanding.

A method for analysing the risk of taking a too low reserve level by use of Chain Ladder method is developed. We give an answer to the question of how much safety loading in terms of the Chain Ladder standard error has to be added to the Chain Ladder reserve in order to reach a specified security level in loss reservin…

We characterize the boundaries of positive holomorphic chains (with both compact and non-compact support) in an arbitrary complex manifold. We then consider a compact oriented real submanifold of dimension 2p-1 in a compact Kahler manifold X and address the question of which relative homology classes in H_{2p}(X,M;Z) a…

New quantum code lacks sparse lift.

Non-negative curvature affects Markov chains' mixing and expansion properties.

Identity testing for reversible Markov chains without symmetry assumption.

Study shows quasipositive fiber surfaces can't be well-ordered.

This thesis is divided into three parts. In the first part, we give an introduction to J. Harrison's theory of differential chains. In the second part, we apply these tools to generalize the Cauchy theorems in complex analysis. Instead of requiring a piecewise smooth path over which to integrate, we can now do so over …

The paper proves inequalities for Steklov eigenvalues on finite graphs.

Markov chains and diffusion processes are indispensable tools in machine learning and statistics that are used for inference, sampling, and modeling. With the growth of large-scale datasets, the computational cost associated with simulating these stochastic processes can be considerable, and many algorithms have been p…

Deep networks are shown to be equivalent to a new type of kernel chain.

The paper shows how to answer future and past questions from high-dimensional time series data.

Classifier chains are popular and effective method to tackle a multi-label classification problem. The aim of this paper is to study the asymptotic properties of the chain model in which the conditional probabilities are of the logistic form. In particular we find conditions on the number of labels and the distribution…

Markov chain Monte Carlo (MCMC) methods are widely used in machine learning. One of the major problems with MCMC is the question of how to design chains that mix fast over the whole state space; in particular, how to select the parameters of an MCMC algorithm. Here we take a different approach and, similarly to paralle…

The paper studies how quickly samples from Langevin dynamics become independent.

FinML-Chain integrates blockchain data for financial machine learning.

The chains studied in this paper generalize Chern-Moser chains for CR structures. They form a distinguished family of one dimensional submanifolds in manifolds endowed with a parabolic contact structure. Both the parabolic contact structure and the system of chains can be equivalently encoded as Cartan geometries (of d…

Transformers with CoT don't enhance reasoning power across all tasks.

This is an expository article on the theory of Kuranishi structure and is based on a series of pdf files we uploaded for the discussion of the google group named `Kuranishi' (with its administrator H. Hofer). There we replied to several questions concerning Kuranishi structure raised by K. Wehrheim. At this stage we su…

A new multi-phase approach improves supply chain forecasting accuracy.

Existence proved for $q$-Bass martingales with specific marginals.

Study non-negative curvature Markov chains, proving entropy contraction.

Training activation quantized neural networks involves minimizing a piecewise constant function whose gradient vanishes almost everywhere, which is undesirable for the standard back-propagation or chain rule. An empirical way around this issue is to use a straight-through estimator (STE) (Bengio et al., 2013) in the ba…

A new stopping rule based on E-values helps efficiently use sampling in Bayesian Deep Ensembles.

Proteins are linear molecular chains that often fold to function. The topology of folding is widely believed to define its properties and function, and knot theory has been applied to study protein structure and its implications. More that 97% of proteins are, however, classified as unknots when intra-chain interaction…

We describe a new class of learning models called memory networks. Memory networks reason with inference components combined with a long-term memory component; they learn how to use these jointly. The long-term memory can be read and written to, with the goal of using it for prediction. We investigate these models in t…

We study two--generated subgroups $\langle f,g\rangle<\mathrm{Homeo}^+(I)$ such that $\langle f^2,g^2\rangle$ is isomorphic to Thompson's group $F$, and such that the supports of $f$ and $g$ form a chain of two intervals. We show that this class contains uncountably many isomorphism types. These include examples with n…

One long-term goal of machine learning research is to produce methods that are applicable to reasoning and natural language, in particular building an intelligent dialogue agent. To measure progress towards that goal, we argue for the usefulness of a set of proxy tasks that evaluate reading comprehension via question a…

Transformers learn chain-of-thought reasoning for longer problems, proving length generalization.

How does supply uncertainty affect the structure of supply chain networks? To answer this question we consider a setting where retailers and suppliers must establish a costly relationship with each other prior to engaging in trade. Suppliers, with uncertain yield, announce wholesale prices, while retailers must decide …

We propose a version of the Hodge conjecture in Bott-Chern cohomology and using results from characterizing real holomorphic chains by real rectifiable currents to provide a proof for this question. We define a Bott-Chern differential cohomology and use atomic section theory of Harvey and Lawson to construct refined Bo…

This work examines robust MCMC for pathological distributions.

This paper studies transformer learning dynamics and initialization.

Study reveals how depth of reasoning affects generalization in models.

Bayesian neural networks show complex posterior distributions that HMC can capture effectively.

The starting point of this article is the question "How to retrieve fingerprints of rhythm in written texts?" We address this problem in the case of Brazilian and European Portuguese. These two dialects of Modern Portuguese share the same lexicon and most of the sentences they produce are superficially identical. Yet t…

The paper explores how LLMs with CoT improve performance on complex tasks.

Transformers can generalize to a large task family with only a few demonstrations.

Study finds on-chain data can proxy off-chain cryptocurrency pricing.

TIM framework uses LLMs and domain experts to infer DeFi user transaction intents.

All parabolic geometries, i.e. Cartan geometries with homogeneous model a real generalized flag manifold, admit highly interesting classes of distinguished curves. The geodesics of a projective class of connections on a manifold, conformal circles on conformal Riemannian manifolds, and Chern--Moser chains on CR--manifo…

Ranking is a key aspect of many applications, such as information retrieval, question answering, ad placement and recommender systems. Learning to rank has the goal of estimating a ranking model automatically from training data. In practical settings, the task often reduces to estimating a rank functional of an object …

C-IP improves LLMs' query selection for interactive tasks by estimating uncertainty robustly.

The study connects monopole chains to Higgs bundles and classifies symmetric chains.

Paper proposes real-time risk metrics for stablecoin protocols.

New proof of chain duality for simplicial complexes.

Improves multi-label classification with a new network model.

Reduces identity testing of reversible Markov chains to simpler symmetric chain tests.